Skip to content

Maintainer Notes

This page is for people extending the plugin itself, not end users running the workflow.

Design rules

  • Keep SKILL.md wrappers lean and put command behavior in references/source-commands/ or references/local-commands/.
  • Put shared rules in references or scripts instead of duplicating them across SKILL.md files.
  • Preserve file-backed handoff contracts between phases.
  • Treat the orchestrator as workflow control, not as a substitute for downstream phase skills.

Development workflow

  1. Edit the relevant SKILL.md, script, or reference file.
  2. Update agents/openai.yaml if user-facing labels or invocation wording changed.
  3. Refresh docs when visible behavior changed.
  4. Run python3 plugins/specark/scripts/validate_plugin_bundle.py.
  5. Run npm run docs:build before publishing docs changes.

Source of truth

Upstream command text lives under plugins/specark/references/source-commands/.

SpecArk-authored local command text lives under plugins/specark/references/local-commands/.

Both forms are authoritative for their wrappers. source-commands/ is reserved for copied upstream originals; local workflow contracts should use local-commands/ instead of embedding long behavior in SKILL.md.

Docs site

Local commands:

bash
npm install
npm run docs:dev
npm run docs:build
npm run docs:preview

GitHub Pages setup (one-time):

  1. Open repository settings → Pages.
  2. Set Build and deployment > Source to GitHub Actions.
  3. Push to main and wait for the deploy workflow to finish.

Notes for further development

  • Keep documentation split between user-facing operation and maintainer-facing release work.
  • Expand skill docs when behavior changes, but avoid turning them into copies of command files.
  • If you broaden test generation beyond API workflows, document the new scope clearly in both the README and the skill pages.

Structured Prompt-Driven Development — works with Codex and Claude Code.